Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS7137039 B2
Publication typeGrant
Application numberUS 10/075,140
Publication dateNov 14, 2006
Filing dateFeb 14, 2002
Priority dateMar 23, 2001
Fee statusPaid
Also published asUS20020138791
Publication number075140, 10075140, US 7137039 B2, US 7137039B2, US-B2-7137039, US7137039 B2, US7137039B2
InventorsPaul Durrant, Stephen R Hanson, David S Gordon, Jeremy Harris
Original AssigneeSun Microsystems, Inc.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Device drivers configured to monitor device status
US 7137039 B2
Abstract
A device driver (GRAPHICS, NETWORK, H2IO, IO2L, SERIAL) for use in a computer system comprising a processor (P), memory (M) and a device (GRAPHICS, NETWORK, H2IO, IO2L, SERIAL) operatively coupled to the computer system, the device driver being operable to control the device to monitor an operational status of the device, and consequent upon a change in the operational status to generate a fault report data indicating whether the change of status was caused internally within the device or externally by another connected device which caused the change of operational status to occur. The fault reports may also include an indication of the operational status of the device. Furthermore, if the fault report data indicates that the change of status was caused externally, the device driver may generate fault direction information indicative of an apparent direction on a connection between the device and the other device suspected as causing the indicated external fault. By providing an indication as to whether a fault has occurred internally or externally, and if externally, a relative direction on a data path from which the fault is perceived, a fault response processor can estimate the location of a faulty device of the computer system with improved accuracy.
Images(10)
Previous page
Next page
Claims(8)
1. A computer system, comprising:
a processor; and
a memory coupled to the processor, wherein the memory comprises program instructions configured to implement:
a plurality of device drivers, each operable to:
monitor an operational status of one of a plurality of devices, wherein to monitor the operational status the device driver is configured to generate environment data representative of at least one parameter value of at least one sensor associated with the device; and
consequent upon a change in the operational status of the monitored device, to generate fault report data indicating whether the change of operational status of the monitored device was caused internally within the monitored device or externally by another device connected to the monitored device;
wherein the plurality of device drivers are all implemented by program instructions from the same memory.
2. A computer system as claimed in claim 1, wherein the fault report data includes an indication of an operational status of the monitored device.
3. A computer system as claimed in claim 1, wherein, if the fault report data indicates that the change of operational status of the monitored device was caused externally, the device driver is operable to generate fault direction information indicative of a connection from which the external fault is perceived.
4. A computer system as claimed in claim 1, wherein the operational status of the monitored device is one of: up, indicating no fault, degraded, indicating that the monitored device is still operational but with impaired performance, or down, indicating that the monitored device is not operational.
5. A computer system as claimed in claim 4, wherein the operational status of the monitored device is determined from at least one of: a time to respond to a command, an amount of data communicated via an I/O bus, an amount of data processed by the monitored device, whether information is being correctly processed, or from an error interrupt signal generated by the monitored device.
6. A computer system as claimed in claim 1, wherein the program instructions are further configured to implement a fault response process operable to analyze generated fault report data generated by one or more of the plurality of device drivers to determine a faulty one of the plurality of devices.
7. A computer system as claimed in claim 1, wherein each of the plurality of device drivers generates the operational status information from at least one of: a number of memory accesses performed, a time taken to respond to a command, and an amount of data processed.
8. A computer system, comprising:
a processor; and
a memory coupled to the processor, wherein the memory comprises program instructions configured to implement:
a plurality of device drivers, each operable to:
monitor an operational status of one of a plurality of devices,
generate operational status information from at least one of: a number of memory accesses performed, a time taken to respond to a command, and an amount of data processed; and
consequent upon a change in the operational status of the monitored device, to generate fault report data indicating whether the change of operational status of the monitored device was caused internally within the monitored device or externally by another device connected to the monitored device;
wherein the plurality of device drivers are all implemented by program instructions from the same memory.
Description
BACKGROUND OF THE INVENTION

The present invention relates to computer systems.

There are many fields in which mankind has become reliant on computers to perform valuable and sometimes essential functions. The reliance on computer systems demands that the down time of computer systems is as small as possible. The down time of a computer system is a period during which a computer system is inoperable as a result of a fault in the system. If a computer system goes down, the inconvenience and loss of revenue and indeed life endangering effects can be substantial. As result, the reliability of computer systems is arranged to be as high as possible.

In a co-pending International patent application serial number US99/124321, corresponding to U.S. patent application Ser. No. 09/097,485, a fault tolerant computer system is disclosed in which multiple processing sets operate to execute substantially the same software, thereby providing a amount of redundant processing. The redundancy provides a facility for detecting faults in the processing sets and for diagnosis and automatically recovering from the detected faults. As a result, an improvement in the reliability of the computer systems is effected, and consequently the down time of such fault tolerant computer systems is likely to be substantially reduced.

Computer systems are generally comprised of a processor and memory connected via an I/O bus to utility devices which serve to provide under control of the processor particular functions. Although redundant processing sets within a computer system provide a facility for detecting, diagnosing and recovering from errors in the processing sets, the utility devices within the computer system, including the connecting buses and peripheral buses, may fail from time to time. A device failure can cause disruption in the operation of the computer system, and may even cause the computer system to go down. Conventionally, detecting and identifying a faulty device or group of devices has required the presence of a skilled technician.

It is therefore desirable to provide a computer system in which a faulty device or group of devices can be readily identified, so that repair can be effected quickly, and down time of the computer system can be reduced.

SUMMARY OF THE INVENTION

Particular and preferred aspects of the invention are set out in the accompanying independent and dependent claims. Combinations of features from the dependent claims may be combined with features of the independent claims as appropriate and not merely as explicitly set out in the claims.

In accordance with one aspect of the invention, there is provided a device driver for use in a computer system comprising a processor, a memory and at least one device, the device driver being operable to monitor an operational status of the device, and consequent upon a change in the operational status, to generate fault report data indicating whether the change of status was caused internally within the device or externally by another connected device.

By providing device drivers which generate fault reports which include an indication of whether a change of operational status was caused internally or externally, a fault response processor can generate automatically, from the fault reports, an estimation of the identity of a device, or a group or groups of devices of the computer system which are likely to be faulty. This is particularly advantageous because another part of the computer system such as an I/O bus may have caused the change in operational status of the device and may in fact be faulty rather than the device itself. In some embodiments therefore, if the fault report data indicates that the change of status was caused externally, the device driver may be operable to generate fault direction information indicative of a connection from which the external fault is perceived. The direction of the external fault therefore provides an indication of which of the devices caused the external fault as perceived by the device driver driving the device.

Although an indication of the relative direction of the fault from the device driver provides an improvement in estimating the location of a faulty device or group of devices, in some embodiments, the fault report information may also include an indication of an operational status of the device. The operational status can therefore be used to improve further the accuracy of the estimated location of the faulty device or group of devices. To this end, in some embodiments the operational status of the device may be indicated as being one of up, indicating no fault, degraded, indicating that the device is still operational but with impaired performance or down indicating that the device is not operational. By applying this operational status information to the model of the devices of the computer system in combination with the relative direction on the data path from which the fault is perceived, the faulty device or group of devices may be located from this information alone without requiring any further analysis.

The device drivers may be arranged to determine an operational status from performance parameters of the devices they are controlling, such as, for example, a time to respond to a command, an amount of data received via an I/O bus.

In some embodiments, the device drivers may be arranged to generate environment data, where the device controlled by the device driver includes a component which is being monitored by a sensor. The term environment data is therefore used to describe any information of parameters, logical flags or signals, which provide information appertaining to the operational status of components which are being monitored within the device. Generating environment data with or separately from the fault reports provides further evidence which can be used to identify the location of a faulty device or group of devices.

An aspect of the invention also provides a computer program providing computer executable instructions, which when loaded onto a computer configures the computer to operate as the device driver. An aspect also includes a computer program product comprising a computer readable medium having recorded thereon information signals representative of the computer program.

The computer readable medium can be any form of carrier medium for carrying computer program code, whether that be a magnetic, optical or any other form of data storage such as a tape, disk, solid state, or other form of storage providing random or read-only or any other form of access, or a transmission medium such as a telephone wire, radio waves, etc.

BRIEF DESCRIPTION OF THE DRAWINGS

Exemplary embodiments of the present invention will be described hereinafter, by way of example only, with reference to the accompanying drawings in which like reference signs relate to like elements and in which:

FIG. 1 is a schematic overview of an example computer system;

FIG. 2 is a schematic block diagram showing the data paths between the drivers for the devices of FIG. 1;

FIGS. 3 is a schematic representation of Field Replaceable Units from which the computer system of FIG. 1 is comprised;

FIG. 4 is a schematic diagram illustrating an inter-relationship of the Field Replaceable Units of FIG. 3;

FIG. 5 is a schematic representation of an Automatic Fault Response processor coupled to the device drivers for the devices of the computer system shown in FIG. 1;

FIG. 6 provides a graphical illustration of a process of identifying analysis intervals (time epochs) used by the Automatic Fault Response processor to analyse fault reports;

FIG. 7 provides a graphical representation of the analysis intervals (time epochs) identified by the Automatic Fault Response processor;

FIGS. 8, 9, and 10, provide example illustrations of an analysis applied by the Automatic Fault Response processor, in which example fault reports are applied to a device tree;

FIG. 11 provides an example of environmental sensors embodied within the Field Replaceable Units forming part of the computer system of FIG. 1;

FIG. 12 provides an illustration of a mapping of the environmental sensors onto a device tree;

FIG. 13 is a flow diagram illustrating the generation of the device tree model representing possibly faulty devices by the Automatic Fault Response processor;

FIG. 14 is a somewhat schematic flow diagram illustrating the operations performed by the Automatic Fault Response processor to identify a FRU containing a faulty device; and

FIG. 15 is a flow diagram illustrating a post-processing operation performed by the Automatic Fault Response processor.

DESCRIPTION OF EMBODIMENTS

Embodiments of the present invention find particular application with a computer system having a plurality of devices which are controlled by device drivers which typically form part of user software executed on a central processor unit. As mentioned above, the devices of the computer system may be divided into groups, each group may contain a single device or a plurality of devices. Each group may be incorporated as part of a Field Replaceable Unit (FRU). A FRU is an item of hardware that can be added to or removed from a system by a customer or by a service engineer at a site where a computer system is installed. A computer system could be constructed entirely of FRUs. A desktop computer system, for example, in its simplest form may have a single FRU on which all the hardware devices of the desktop computer are incorporated, except for a monitor and keyboard which may form separate FRUs. A server however may be constructed of many FRUs: motherboards, CPU sets, peripherals, disks, for example which are inter connected.

The FRUs of a computer system will typically have an inter-dependent hierarchy, which is related to a hierarchical inter-relationship of the devices of the computer system, although a FRU may contain more than one device and so there may not be a direct correspondence between the device hierarchy and the FRU hierarchy.

Within a computer system the kernel software arranges the devices of the system in accordance with the device hierarchy. An example of a computer system with which the present invention finds application is shown in FIG. 1. In FIG. 1 a CPU 2, and memory 4 are connected to a host bus H. Also connected to the host bus H is a host bus to IO bus bridge H2IO and a graphics device 6. The host to IO bus bridge is connected to a main I/O bus IO, to which is connected a network device 8 and an IO to L bridge IO2L. The network device is representative of a medium bandwidth device. A slow device 12 such as one which may be operating a serial interface is connected to the IO to L bridge via a low bandwidth bus L.

The devices of the computer system shown in FIG. 1 can be regarded as forming a hierarchical tree structure. At the root of the hierarchy is a node representing the host bus H of the system, which is the bus to which the CPU 2, and memory 4 are connected. Nodes for peripheral devices such as Ethernet chips, and serial UARTs form the leaves of the tree structure which are attached below nodes representative of the buses to which these devices are attached. The device tree structure for the computer system of FIG. 1 is shown in FIG. 2. The device tree shown in FIG. 2 represents the data paths between the drivers of the devices of the computer system of FIG. 1. Each device in the hierarchy has an associated device driver.

The devices shown in FIG. 2 will be incorporated within one or more FRUs. The computer system is constructed from these FRUs. FIG. 3 provides an example mapping of the devices of the computer system shown in FIG. 1 on to four FRUs 20, 22, 24, 26. A first FRU 20 forms a motherboard of the computer system, a second FRU forms a graphics device 26, a third FRU 22 forms a network interface and a fourth FRU 24 forms a serial interface 24. In accordance with the device hierarchy shown in FIG. 2, a relative dependency will be formed upon the FRUs of the computer system. Accordingly for the present example, this is illustrated in FIG. 4, where the FRU structure shown in FIG. 3 is illustrated with a relative dependency illustrated by arrows 30.

Generally, the relative dependent relationship between the FRUs and the mapping between FRU and device hierarchies is maintained in a library file.

Embodiments of the present invention provide a facility for readily identifying a FRU of the computer system which contains a device which has developed a fault. A FRU which contains a device which has developed a fault will be referred to in the following description as a faulty FRU. This is provided by an Automatic Fault Response processor. The faulty device may be for example one of the peripheral devices, but may also be one of the connecting buses H, H2IO, IO2L. As will be explained shortly, the most likely FRU which contains the faulty device is identified from fault reports generated by device drivers within the kernel software running on the CPU.

Device Drivers

In FIG. 5 the AFR is shown generally to be connected through bi-directional links 50 to the device drivers GRAPHICS, NETWORK, H2IO, IO2L, SERIAL. Each of the device drivers is arranged to control the operation of a corresponding one of the devices which are represented in FIG. 2.

The AFR processor is also shown communicating with a library file 80 of the kernel. The library file 80 provides the relative dependent relationships between the FRUs and the mapping between the FRUs and the device hierarchy.

The device drivers are able to monitor the operation of the devices, through for example, a time for responding to commands, an amount of data processed by the device, a number of memory accesses, whether information is being correctly processed and other measures of relative performance. The devices are therefore able to detect a change in relative performance for the device.

Each of the device drivers GRAPHICS , NETWORK, H2IO, IO2L, SERIAL determines the operational status of the device. When there is a change in operational status a fault report is generated. In one example embodiment, the fault reports have the following fields:

    • Device datapath e.g./H2IO/IO2L/SERIAL
    • New state=down, degraded or up
    • Location=data path fault, device fault(internal) or external fault.

As will be explained shortly, the fault reports generated by the device drivers are used by the AFR processor to identify the FRU or FRUs of the computer system which contain a faulty device, referred to as a faulty FRU. However, in addition to the fault reports, the AFR utilises information from environment sensors. These may form part of the devices within the FRUs or may not be associated with any one device but rather monitor environmental conditions on the FRU as a whole. The sensors provide data which are representative of the values of the sensed parameters provided by the sensors for generating environmental information. The environmental information provides an indication of the operating status of components within the devices with respect to where the sensors are located. The sensed parameters may be for example, temperature, power consumption, or fan speed.

A separate management driver may be provided to interrogate the sensors or to retrieve data produced by the sensor from a cached memory. The management driver may then communicate the environment data to the AFR. Alternatively, the device driver for a device may retrieve the environment data from a sensor associated with the device driver and pass this to the AFR.

Automatic Fault Response Processor

The operation of the AFR processor to detect a faulty FRU from fault reports generated by device drivers will now be explained. The AFR performs two main functions. The first function is to extract information from the fault reports generated by the device drivers and to log the fault reports. The second function is to perform a diagnosis so as to estimate which of the FRUs of the computer system is or are faulty. To do this the AFR first builds a device tree in its own data space. This is represented by FIG. 5 as device tree DT constructed within the data space DS of the AFR. The device tree is constructed by adding or updating nodes for all the devices featured in the data paths of the fault reports collected during a period of time, called an epoch. The tree so constructed is hence not necessarily a complete copy of the full kernel device tree. A new tree is built during each new epoch. A detailed explanation of the formation of epochs will be provided shortly.

In some embodiments, the device tree is built by the AFR as follows:

For each fault report, extract the device data path and use it to search down the current tree for a device node. If no such node exists, create a new one with a state of UP and empty fault location information. If a node does exist, update it according to the following rules:

    • If the device node state is UP then the location information in the fault report is considered to be the most significant indication of the location of the faulty device. The information is therefore copied into the node in the tree.
    • If the device node is DEGRADED and the fault report declares service to be LOST, the node state is changed to DOWN and the location information from the fault report is considered to be the most significant fault location.
    • If the device node is DEGRADED and the fault report declares service to be DEGRADED, or the device node is DOWN and the fault report declares service to be LOST, then the location information from the fault report is considered to be more significant if it indicates a fault higher up the device tree i.e. DATAPATH is more significant then DEVICE, and the DEVICE is more significant than EXTERNAL.

If the fault report declares service to be RESTORED then any location information is cleared from the device node and its state is changed to UP.

The model of the device tree forms the basis of the AFR's analysis. Analysis is performed in three phases. The purpose of the phases is, if possible, to identify the faulty FRU, with each analysis phase providing a further refinement in the estimation of the identity of the faulty FRU. As will be explained shortly, this is effected by assigning a fault probability to the FRU containing some or all of the devices in the device tree and declaring a FRU as faulty if it has a probability which exceeds a predetermined threshold. The three phases will be explained shortly. The formation of the time epochs will now be described.

Time Epochs and Dynamic Tree Building

As explained above the fault reports are analysed within an analysis interval which is referred to in the following description as a ‘time epoch’. Time epochs are determined from a rate of arrival of the fault reports. This is because the fault reports generated by the device drivers can be correlated. As such, although only a single device may be faulty, other devices may experience a change in operational status so that several fault reports are generated. As a result, the fault reports may be related, and the relationship may be reflected in a time at which the fault reports are generated. The fault reporting can have, therefore, a certain periodicity as a result of specific operations being performed by that device or an access being made to that device. By identifying, according to this periodicity, a time epoch corresponding to a generation cycle of the fault reports, an improvement in the likelihood of correctly locating the faulty device can be provided. This is represented schematically in FIG. 6.

In FIG. 6 the horizontal lines 90, 92 represent the passage of time going from left to right across the page. Each of the boxes 94 between the horizontal line 90, 92 represents a period during which a fault report or reports may be generated. The fault reports are analysed as mentioned above and used to update the device tree. The reference period is used to identify whether there has been sufficient recent fault report activity to indicate a change in the status of the devices to form the start GO and end STOP a time epoch 164. The reference periods are referred to as ticks and are shown in FIG. 7 which provides a graphical illustration of a number of device node changes in each tick with respect to time.

In order to identify a time epoch, the AFR monitors the device tree model to determine how many new nodes have been added to the device tree model or how many existing nodes have been updated since the previous tick. If there have been no changes since the previous tick but activity on the model has occurred, then an end of epoch is declared, and analysis on the device tree is performed. If there was no quiet period, which corresponds to a tick where there were no changes to the tree, in the last n ticks, then the tick period, T, is halved so that shorter quiet periods can be analysed. The graphical representation provided in FIG. 7 illustrates an analysis period between ends of epochs D_EPCH, and a period RTCK following n ticks without a quiet period ACT, in which the tick period is halved. The time epochs are identified from the rate of arrival of fault reports, and the changes that these fault reports make to the device tree model.

An epoch may continue indefinitely until the device tree changes. Once a change has occurred however, the maximum remaining epoch time as the tick period is progressively halved can be expressed generally by the following expression:
nT+nT/2+nT/4+nT/8+ . . . →2nT

There is however one exception to this bound on the epoch length. A time epoch which begins at the start of a boot configuration period of the computer system will continue until the boot configuration has been completed. The AFR processor operates in a first phase, as explained above to identify the time epoch within which it is assumed that the fault reports are correlated. The fault reports for the time epoch are collected and used to construct a device tree 81 by ‘decorating’ the nodes with the current status of the devices. Operational states are represented by updating the current state of the node with information from the fault report according to the rules given above. The tree structure allows faults to be related hierarchically. Analysis modules of the AFR may use this information to modify the information on the tree.

Analysis Phases

Having built a device tree representing the possibly faulty devices, the AFR proceeds to analyse the device tree in order to attempt to identify the faulty FRU which contains the faulty device. This is done in three phases:

Phase I

The AFR performs phase I by modifying the nodes of the device tree, which was built during the time epoch to eliminate redundant fault reports. This is achieved in accordance with a set of rules. For example, if a parent node indicates a device fault, any fault indicated by a child node may be a false positive and so it may be desirable to clear the fault information of these child nodes. Effectively, the AFR processor is preprocessing the device tree in order to remove any fault reports which are redundant.

EXAMPLE

FIG. 8 shows that the drivers for both devices A and C have positively identified their device as having a fault, FR.1, FR.2. In this case the evidence from the driver for device C, FR.2 is discounted, because the fault report was likely to have been triggered as a result of reading bad data through A, although this could not be determined at the time the fault was reported.

Phase II

In the second phase of the operation the device tree is analysed by the AFR to identify a set of faulty FRUs with a non-zero probability of having a fault. For example, if a device node is down and indicating that it is the location of the fault then there is a 100% probability that the FRU containing that device has a fault. If a device node is down and is indicating a fault in its data path and an ancestor is indicating an external fault then the fault is deemed to lie in a FRU containing either of the two devices or a FRU in between (if there is one). Hence a 100% probability is assigned to a set of FRUs but not to an individual.

In some embodiments the AFR is provided with a plurality of analysis modules Mn each of which implements a single type of Phase I, Phase II or Phase III (see below) analysis. In Phase II, for each FRU, each module Mn (that implements a Phase II type of analysis) assigns a (possibly zero) probability Pn that there is a fault on that FRU. The Modules can assign probabilities to more than one FRU. If a FRU receives a non-zero probability of being faulty from more than one module, then the probabilities are combined as follows:
(1−P)=(1−P 1)(1−P 2) . . . (1−P n)

Therefore the probability that a particular FRU is not faulty is the probability that all the modules determine that it is not at fault. After phase III analysis has been performed which will be described shortly, the probability for each FRU is compared with a threshold and if greater than the threshold, then the FRU or FRUs are declared as being faulty.

The following examples provide a further illustration of the analysis of the device tree 81, to identify a possibly faulty set of FRUs:

Example A

Consider the AFR constructed device tree in FIG. 9. The driver for device A has reported an external fault FR.3 and the driver for device C has positively identified an internal fault FR.4. The device C is unambiguously identified as being in error (P=100%). The FRU containing this device is therefore considered to be faulty.

Example B

FIG. 10 shows that the driver for device A has reported an external fault FR.5 and the driver for device C have reported a data path fault FR.6. The analysis modules form a probability metric that one of the FRUs contains a faulty device, or that the fault lies somewhere between devices A and C (possibly including the devices themselves). In this case the fault probability that a FRU contains a faulty device is weighted between the number of devices on the FRU. For the present example, if the devices A, C are embodied on the same FRU, then this FRU is assigned a 100% fault probability. If however the two devices are embodied on different FRUs then each FRU is assigned a fault probability of 50%. However, if the fault probability metric generated does not exceed the predetermined probability threshold then no conclusion may be drawn as to the faulty FRU. An improved estimate of the identity of the faulty FRU can be made from analysis performed in accordance with phase III.

Phase III

In a third phase of the operation of the AFR, the list of possibly faulty FRUs from phase II is examined further by applying environmental information provided by appropriate sensors. The information from the sensors from each FRU is checked for environmental problems such as power-loss, over-temperature, etc. This information is used to adjust the fault probabilities of the FRUs.

As illustrated in the examples given above, in some circumstances, the fault report information may not be conclusive and so the estimate of the faulty FRU may only identify a plurality of FRUs which may be faulty. For this reason the phase III analysis is operable to apply environmental reports to the device tree in order, if possible, to produce an improved probability estimate of the faulty FRU.

An example configuration of FRUs is shown in FIG. 11, with the associated device tree shown in FIG. 12. As shown in FIG. 11 the example FRUs are a mother board MBD, a slot SLT and a network card NET which are connected to the mother board FRU. The device tree shown in FIG. 12, also includes environment sensors, which provide sensed parameters relating to temperature TEMP and fan speed FAN.

In the third phase of the analysis, environmental information provided by the sensors TEMP, FAN from the FRUs is applied to the FRU list. In order to reduce the likelihood of false data being provided from the environmental information, a sensor device path may be used to determine whether the sensor device itself is down, in which case the environmental information is disregarded. The AFR processor uses the environment information to improve the estimate of faulty FRUs which resulted from phase II analysis. Where for example, the phase II analysis identifies only a group of FRUs which may be faulty, the environment data can be used to provide a more accurate estimate of the faulty FRU, by selecting a FRU having an abnormal sensor reading. Again, even after the environment information has been applied, it is possible that the estimate of the faulty FRU only identifies a group of FRUs. However it may be sufficient that enough report information has been acquired to identify that one or more FRUs within a group of FRUs are suspected as being at fault. This information is therefore valuable to a technician assigned to repair the computer system and to this end this information is generated with the fault reports on a graphical user interface to be accessed by the technician assigned to repair the computer system.

When all three phases of analysis are complete, the resultant list of FRU fault probabilities is examined and compared against a threshold value, for example, 90%. Any FRU having a fault probability in excess of this is deemed to have failed. The AFR indicates that a FRU is faulty, by marking the FRU as such. In some embodiments, a message is generated by the AFR, which is written to a non-volatile storage on the faulty FRU. The faulty FRU may be indicated by causing an LED to illuminate on the FRU. The operation of the post-analysis phase will now be explained in more detail.

Post Analysis Phase

If a FRU can be positively identified as being faulty, a repair technician can be alerted to this fact and to this end the AFR processor may signal that a FRU is faulty through an interface which is used to change the state of the FRU to faulty. This may be notified to a repair and maintenance organisation. Furthermore the FRU may carry a ‘change me’ LED so that the FRU can be easily identified by a technician. Alternatively, where a group of FRUs are suspected as being faulty, then each can be signalled as being possibly faulty. Accordingly, it will be appreciated that there are various ways for providing an external signal to indicate that a FRU is faulty, to a technician. Furthermore, the fault diagnosis may be written into a non-volatile storage medium on a board into which the FRU is loaded to aid diagnosis when the FRU is repaired.

In summary, in the three phases the AFR processor combines more than one device fault report and/or environmental information reports from different parts of a computer system and automatically determines the most likely area of the system where a faulty FRU or device or group of devices is located and the devices which are affected by the fault. If there is sufficient evidence then one of the FRUs of the computer system may be declared faulty. This provides both an automated and an early recognition of devices which are faulty which can be used by a system administrator to initiate repairs before a device has completely failed.

Summary of Operation

The operation of the AFR processor is summarised in the form of flow diagrams which are shown in FIGS. 13, 14 and 15. FIG. 13 provides a flow diagram which illustrates the operation phase of the AFR processor when identifying the time epochs and building the device tree, before the analysis process is performed as shown in FIG. 14.

FIG. 13 illustrates the process through which the time epochs are identified. The process starts at process step 300, following which the AFR processor receives the fault reports generated by the device drivers at step 302 for the current tick period, and the ‘tick’ advanced. This forms part of a first pre-analysis phase. At process step 304 the AFR processor uses the information provided by the fault reports to build a device tree, by adding devices to the tree which are indicated or suspected as being possibly faulty or which detect a fault. At decision step 306, it is determined whether the device tree has changed from the previous epoch. If the device tree has not changed, then an end of epoch is declared at process step 310, and at process step 311, the analysis phase is performed and ends at step 312.

If the device tree has changed, a further decision step 308 is provided in order to determine whether or not it is necessary to adjust the tick period. If the device tree has changed for n consecutive tick periods, then the tick period is adjusted to the effect of making the tick period shorter, so that a temporal resolution of the analysis performed with respect to the tick periods is better matched to the arrival rate of the fault reports. If the device tree has changed for n consecutive periods, then the tick period is adjusted at step 314. Otherwise this step is bypassed.

The analysis process is represented by the flow diagram shown in FIG. 14. In FIG. 14, the pre-analysis process of generating the device tree DT from the fault report information collected in a time epoch is represented generally as the step 400. The device tree DT representing the possibly faulty devices is shown as an input to the first analysis phase P.1. The AFR includes a plurality of analysis modules Mn each of the modules being provided for a particular type of analysis, as mentioned above. The analysis modules of the AFR perform the phase 1 analysis by removing fault reports which will not be helpful in identifying the faulty FRU according to the set of rules explained above. Following the phase 1 process P.1, an adjusted device tree DT′ is provided as an input to the second phase of the analysis P.2. During the phase 2 analysis P.2, the fault probability of the FRU containing the devices in the device tree DT′ is determined from the fault report information provided for the devices in the device tree DT′. Each module Mn is operable to calculate the probability Pn of a FRU being faulty, from information generated by a device embodied within the FRU. At this point one or more FRUs FRUA, FRUB may be identified as possibly being faulty. However during phase three P.3, the environment information is applied to the identified FRUs, in order to refine the estimate of which of the FRUs FRUA, FRUB is faulty. As indicated above in some embodiments, this is effected by identifying whether any of the FRUs FRUA, FRUB returns environment data which indicates an abnormal reading. A threshold probability is then applied and, if any FRU's fault probability exceeds the threshold, this FRU is then declared as being faulty. After the faulty FRU has been identified, the post analysis phase 402 is performed.

The post analysis phase is described by the flow diagram shown in FIG. 15. As shown in FIG. 15, the post analysis phase starts at node 402 and begins with a decision step 322, at which it is determined whether the faulty FRU has been unambiguously located. If the FRU has been unambiguously located, then external signals associated with the FRU or group of FRUs identified as being faulty is or are activated at step 326, to provide an indication to a technician that these FRUs are faulty. Whether or not the faulty FRU or FRUs have not been unambiguously identified, a fault diagnostic report is generated at step 328 which indicates a plurality of possibly faulty FRUs. The fault diagnostic report is then displayed at step 330 on a graphical user interface or communicated to a remotely located site at which appropriate action can be taken to either replace all of the suspected faulty FRUs or to allow a technician to analyse the fault reports and/or environmental data. At this point the process then terminates at step 332.

It will be appreciated that although particular embodiments of the invention have been described, many modifications/additions and/or substitutions may be made within scope of the present invention as defined in the appended claims.

Furthermore various modifications may be made to the embodiments of the invention herein before described without departing from the scope of the present invention. In particular, it will be appreciated that the embodiments of the invention can be applied to any form of computer system in which the computer system is comprised of a plurality of utility devices connected to a kernel comprising a processor and memory on which the kernel software is executed. Furthermore it will be appreciated that either the environmental analysis process corresponding to phase III of the Automatic Fault Response process could be performed separately and distinct from phase I and II of the process in which the fault report information is applied to the device tree and a list of FRUs generated, respectively. More particularly, in other embodiments of the present invention the devices of the computer system may not be embodied within FRUs. In such embodiments the AFR will be operable to identify the device which is most likely to be faulty or a group of devices, from one of several groups into which the devices of the computer system are divided.

In other embodiments, fault reports may be discounted in accordance with predetermined rules, when building the device tree. If, for example, a device is identified from past fault reports as being likely to have an intermittent fault, then this information can be used to discount fault reports associated with this or other devices. Furthermore field engineers could write modules to discount information from specific devices that are suspected as misbehaving at customer sites so providing a combination of automated fault report discounting and additional overriding fault report discounting introduced by the field engineer.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US4554661 *Oct 31, 1983Nov 19, 1985Burroughs CorporationGeneralized fault reporting system
US4649515Jul 1, 1986Mar 10, 1987Westinghouse Electric Corp.Methods and apparatus for system fault diagnosis and control
US5109486Jan 6, 1989Apr 28, 1992Motorola, Inc.Distributed computer system with network and resource status monitoring
US5293556Jul 29, 1991Mar 8, 1994Storage Technology CorporationKnowledge based field replaceable unit management
US5394543 *Mar 30, 1993Feb 28, 1995Storage Technology CorporationKnowledge based machine initiated maintenance system
US5561760 *Sep 22, 1994Oct 1, 1996International Business Machines CorporationSystem for localizing field replaceable unit failures employing automated isolation procedures and weighted fault probability encoding
US5587930Jul 24, 1991Dec 24, 1996Mitsubishi Denki Kabushiki KaishaFault diagnosis device
US5655068 *Jan 17, 1996Aug 5, 1997Adc Telecommunications, Inc.Point-to-multipoint performance monitoring and failure isolation system
US5838899 *Jun 5, 1996Nov 17, 1998Stratus ComputerDigital data processing methods and apparatus for fault isolation
US6000040 *Oct 29, 1996Dec 7, 1999Compaq Computer CorporationMethod and apparatus for diagnosing fault states in a computer system
US6349335Jan 8, 1999Feb 19, 2002International Business Machines CorporationComputer system, program product and method for monitoring the operational status of a computer
US6385665 *Dec 18, 1998May 7, 2002Alcatel Usa Sourcing, L.P.System and method for managing faults in a data transmission system
US6523140 *Oct 7, 1999Feb 18, 2003International Business Machines CorporationComputer system error recovery and fault isolation
US6532552 *Sep 9, 1999Mar 11, 2003International Business Machines CorporationMethod and system for performing problem determination procedures in hierarchically organized computer systems
US6557121 *Mar 31, 1997Apr 29, 2003International Business Machines CorporationMethod and system for fault isolation for PCI bus errors
US6594774Sep 7, 1999Jul 15, 2003Microsoft CorporationMethod and apparatus for monitoring computer system objects to improve system reliability
US6678839 *Jan 31, 2001Jan 13, 2004Nec CorporationTroubleshooting method of looped interface and system provided with troubleshooting function
US6697962Oct 20, 2000Feb 24, 2004Unisys CorporationRemote computer system monitoring and diagnostic board
US20010003835May 11, 1998Jun 14, 2001Watts LavaughnComputer system with environmental detection
US20020138782Mar 8, 2002Sep 26, 2002Paul DurrantComputer system
US20040103181Nov 27, 2002May 27, 2004Chambliss David DardenSystem and method for managing the performance of a computer system based on operational characteristics of the system components
JPH05151707A Title not available
JPH11194961A Title not available
JPS61122759A Title not available
Non-Patent Citations
Reference
1International Search Report application No. GB0107370.9 dated Dec. 11, 2001.
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US7401259 *Jun 19, 2003Jul 15, 2008Sun Microsystems, Inc.System and method for scenario generation in a distributed system
US7774656 *Jan 7, 2009Aug 10, 2010International Business Machines CorporationSystem and article of manufacture for handling a fabric failure
US7966392 *Jul 28, 2006Jun 21, 2011Cyber Solutions Inc.Network management system and network management information collecting method
US8549361 *Dec 21, 2010Oct 1, 2013Netapp, Inc.System and method for construction, fault isolation, and recovery of cabling topology in a storage area network
US20120023583 *Jul 20, 2010Jan 26, 2012Mcafee, Inc.System and method for proactive detection of malware device drivers via kernel forensic behavioral monitoring and a back-end reputation system
US20120159252 *Dec 21, 2010Jun 21, 2012Britto RossarioSystem and method for construction, fault isolation, and recovery of cabling topology in a storage area network
Classifications
U.S. Classification714/47.2, 714/E11.026, 714/44
International ClassificationG06F11/07, G06F11/00
Cooperative ClassificationG06F11/0772, G06F11/079, G06F11/0745
European ClassificationG06F11/07P1K, G06F11/07P4B, G06F11/07P6
Legal Events
DateCodeEventDescription
Apr 16, 2014FPAYFee payment
Year of fee payment: 8
May 3, 2010FPAYFee payment
Year of fee payment: 4
Feb 14, 2002ASAssignment
Owner name: SUN MICROSYSTEMS, INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:DURRANT, PAUL;HANSON, STEPHEN;GORDON, DAVID;AND OTHERS;REEL/FRAME:012617/0197
Effective date: 20010322